| { | |
| "best_metric": 10.252269744873047, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.4548038658328596, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0022740193291642978, | |
| "grad_norm": 0.6463788151741028, | |
| "learning_rate": 1e-05, | |
| "loss": 10.3614, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0022740193291642978, | |
| "eval_loss": 10.362320899963379, | |
| "eval_runtime": 2.1722, | |
| "eval_samples_per_second": 341.127, | |
| "eval_steps_per_second": 85.627, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0045480386583285955, | |
| "grad_norm": 0.6392422914505005, | |
| "learning_rate": 2e-05, | |
| "loss": 10.3628, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006822057987492893, | |
| "grad_norm": 0.6401531100273132, | |
| "learning_rate": 3e-05, | |
| "loss": 10.3624, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.009096077316657191, | |
| "grad_norm": 0.6498253345489502, | |
| "learning_rate": 4e-05, | |
| "loss": 10.3615, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01137009664582149, | |
| "grad_norm": 0.6230224370956421, | |
| "learning_rate": 5e-05, | |
| "loss": 10.3592, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.013644115974985787, | |
| "grad_norm": 0.6378709673881531, | |
| "learning_rate": 6e-05, | |
| "loss": 10.3579, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.015918135304150087, | |
| "grad_norm": 0.6344523429870605, | |
| "learning_rate": 7e-05, | |
| "loss": 10.3554, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.018192154633314382, | |
| "grad_norm": 0.613169252872467, | |
| "learning_rate": 8e-05, | |
| "loss": 10.3548, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02046617396247868, | |
| "grad_norm": 0.6090683341026306, | |
| "learning_rate": 9e-05, | |
| "loss": 10.3537, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02274019329164298, | |
| "grad_norm": 0.6269639134407043, | |
| "learning_rate": 0.0001, | |
| "loss": 10.3495, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.025014212620807278, | |
| "grad_norm": 0.6119576096534729, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 10.3463, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.027288231949971573, | |
| "grad_norm": 0.6316317319869995, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 10.343, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.029562251279135872, | |
| "grad_norm": 0.6255670785903931, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 10.3399, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.031836270608300174, | |
| "grad_norm": 0.6053351759910583, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 10.3382, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03411028993746447, | |
| "grad_norm": 0.595924973487854, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 10.3349, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.036384309266628764, | |
| "grad_norm": 0.6037504076957703, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 10.3319, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.038658328595793066, | |
| "grad_norm": 0.5684234499931335, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 10.3279, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.04093234792495736, | |
| "grad_norm": 0.554912269115448, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 10.3257, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04320636725412166, | |
| "grad_norm": 0.5255692601203918, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 10.3226, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04548038658328596, | |
| "grad_norm": 0.5183072090148926, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 10.3206, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.047754405912450254, | |
| "grad_norm": 0.47872552275657654, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 10.318, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.050028425241614556, | |
| "grad_norm": 0.4611729681491852, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 10.3145, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05230244457077885, | |
| "grad_norm": 0.4263930916786194, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 10.3124, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.054576463899943146, | |
| "grad_norm": 0.424592524766922, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 10.3115, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05685048322910745, | |
| "grad_norm": 0.3940682113170624, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 10.3077, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.059124502558271744, | |
| "grad_norm": 0.37290531396865845, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 10.3068, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.061398521887436046, | |
| "grad_norm": 0.347023606300354, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 10.3047, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06367254121660035, | |
| "grad_norm": 0.326589971780777, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 10.3028, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06594656054576464, | |
| "grad_norm": 0.31014037132263184, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 10.3013, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06822057987492894, | |
| "grad_norm": 0.28567907214164734, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 10.3006, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07049459920409323, | |
| "grad_norm": 0.2634374797344208, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.2978, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07276861853325753, | |
| "grad_norm": 0.24076446890830994, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 10.298, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07504263786242182, | |
| "grad_norm": 0.23588111996650696, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 10.298, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.07731665719158613, | |
| "grad_norm": 0.2201862335205078, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 10.2956, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07959067652075043, | |
| "grad_norm": 0.21873369812965393, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 10.295, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08186469584991472, | |
| "grad_norm": 0.19085676968097687, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 10.2922, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08413871517907902, | |
| "grad_norm": 0.18656592071056366, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 10.2932, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.08641273450824331, | |
| "grad_norm": 0.1768234372138977, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 10.2919, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.08868675383740762, | |
| "grad_norm": 0.19042670726776123, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 10.2922, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.09096077316657192, | |
| "grad_norm": 0.16703151166439056, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 10.2913, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09323479249573621, | |
| "grad_norm": 0.18089061975479126, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 10.2905, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09550881182490051, | |
| "grad_norm": 0.175005242228508, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 10.2887, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0977828311540648, | |
| "grad_norm": 0.1778019666671753, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 10.2921, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.10005685048322911, | |
| "grad_norm": 0.19467297196388245, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 10.2887, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.10233086981239341, | |
| "grad_norm": 0.22236675024032593, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 10.289, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1046048891415577, | |
| "grad_norm": 0.22028876841068268, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 10.2881, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.106878908470722, | |
| "grad_norm": 0.2525130808353424, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 10.291, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.10915292779988629, | |
| "grad_norm": 0.21920563280582428, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 10.2911, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1114269471290506, | |
| "grad_norm": 0.2570950984954834, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 10.2893, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1137009664582149, | |
| "grad_norm": 0.27705293893814087, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 10.2924, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1137009664582149, | |
| "eval_loss": 10.279383659362793, | |
| "eval_runtime": 2.1726, | |
| "eval_samples_per_second": 341.071, | |
| "eval_steps_per_second": 85.613, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11597498578737919, | |
| "grad_norm": 0.2824695408344269, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 10.2785, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.11824900511654349, | |
| "grad_norm": 0.27513623237609863, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 10.2767, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.12052302444570778, | |
| "grad_norm": 0.29252728819847107, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 10.2742, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12279704377487209, | |
| "grad_norm": 0.2675190567970276, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 10.2751, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12507106310403637, | |
| "grad_norm": 0.27211812138557434, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 10.2738, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1273450824332007, | |
| "grad_norm": 0.26511716842651367, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 10.2717, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.129619101762365, | |
| "grad_norm": 0.2620036005973816, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 10.2717, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1318931210915293, | |
| "grad_norm": 0.2654663324356079, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 10.2668, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13416714042069358, | |
| "grad_norm": 0.23434790968894958, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 10.2669, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.13644115974985788, | |
| "grad_norm": 0.24271494150161743, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 10.2662, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13871517907902217, | |
| "grad_norm": 0.27789735794067383, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 10.266, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.14098919840818647, | |
| "grad_norm": 0.25908592343330383, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 10.2631, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14326321773735076, | |
| "grad_norm": 0.2550159692764282, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 10.2636, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14553723706651506, | |
| "grad_norm": 0.20187436044216156, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 10.2609, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.14781125639567935, | |
| "grad_norm": 0.18251566588878632, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 10.2607, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15008527572484365, | |
| "grad_norm": 0.1462724357843399, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 10.2587, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.15235929505400797, | |
| "grad_norm": 0.1344972401857376, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 10.2593, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.15463331438317227, | |
| "grad_norm": 0.13006775081157684, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 10.2602, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.15690733371233656, | |
| "grad_norm": 0.13522356748580933, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 10.2608, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.15918135304150086, | |
| "grad_norm": 0.15374834835529327, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 10.2597, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16145537237066515, | |
| "grad_norm": 0.15410751104354858, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 10.2594, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16372939169982945, | |
| "grad_norm": 0.16489768028259277, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 10.2577, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.16600341102899374, | |
| "grad_norm": 0.1631012260913849, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 10.2607, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.16827743035815804, | |
| "grad_norm": 0.1335574984550476, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 10.2587, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.17055144968732233, | |
| "grad_norm": 0.13278402388095856, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 10.2595, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.17282546901648663, | |
| "grad_norm": 0.13010838627815247, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 10.2567, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.17509948834565095, | |
| "grad_norm": 0.13910891115665436, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 10.2576, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.17737350767481525, | |
| "grad_norm": 0.13217619061470032, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 10.2573, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.17964752700397954, | |
| "grad_norm": 0.12787939608097076, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 10.2591, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.18192154633314384, | |
| "grad_norm": 0.16071626543998718, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 10.2588, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18419556566230813, | |
| "grad_norm": 0.11572972685098648, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 10.2566, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.18646958499147243, | |
| "grad_norm": 0.09892506152391434, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 10.2571, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.18874360432063672, | |
| "grad_norm": 0.10262473672628403, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 10.2563, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.19101762364980102, | |
| "grad_norm": 0.10136529058218002, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 10.2605, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1932916429789653, | |
| "grad_norm": 0.08113791793584824, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 10.2582, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1955656623081296, | |
| "grad_norm": 0.08277690410614014, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 10.2586, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.19783968163729393, | |
| "grad_norm": 0.07317834347486496, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 10.258, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.20011370096645822, | |
| "grad_norm": 0.09127932786941528, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 10.2625, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.20238772029562252, | |
| "grad_norm": 0.08510231971740723, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 10.2611, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.20466173962478681, | |
| "grad_norm": 0.09062691032886505, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 10.2612, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2069357589539511, | |
| "grad_norm": 0.060250334441661835, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 10.2613, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2092097782831154, | |
| "grad_norm": 0.0716901421546936, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 10.2617, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2114837976122797, | |
| "grad_norm": 0.09290280938148499, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 10.2602, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.213757816941444, | |
| "grad_norm": 0.0716087818145752, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 10.2617, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2160318362706083, | |
| "grad_norm": 0.08726319670677185, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 10.2614, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.21830585559977259, | |
| "grad_norm": 0.0775885283946991, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 10.2652, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2205798749289369, | |
| "grad_norm": 0.09004725515842438, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 10.2631, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.2228538942581012, | |
| "grad_norm": 0.11865659058094025, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 10.2714, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.2251279135872655, | |
| "grad_norm": 0.12341362237930298, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 10.2687, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2274019329164298, | |
| "grad_norm": 0.17424803972244263, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 10.2674, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2274019329164298, | |
| "eval_loss": 10.256570816040039, | |
| "eval_runtime": 2.1634, | |
| "eval_samples_per_second": 342.517, | |
| "eval_steps_per_second": 85.976, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2296759522455941, | |
| "grad_norm": 0.07750660181045532, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 10.2543, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.23194997157475838, | |
| "grad_norm": 0.07942042499780655, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 10.2535, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.23422399090392268, | |
| "grad_norm": 0.081281878054142, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 10.2545, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.23649801023308697, | |
| "grad_norm": 0.06754318624734879, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 10.2544, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.23877202956225127, | |
| "grad_norm": 0.06064874306321144, | |
| "learning_rate": 5e-05, | |
| "loss": 10.2538, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24104604889141557, | |
| "grad_norm": 0.06486661732196808, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 10.2554, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.2433200682205799, | |
| "grad_norm": 0.08350766450166702, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 10.2531, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.24559408754974418, | |
| "grad_norm": 0.07456638664007187, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 10.2535, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.24786810687890848, | |
| "grad_norm": 0.06450087577104568, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 10.2538, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.25014212620807275, | |
| "grad_norm": 0.06120529770851135, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 10.2522, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25241614553723707, | |
| "grad_norm": 0.081166572868824, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 10.2536, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.2546901648664014, | |
| "grad_norm": 0.06143665313720703, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 10.2518, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.25696418419556566, | |
| "grad_norm": 0.06174364686012268, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 10.2517, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.25923820352473, | |
| "grad_norm": 0.06238188594579697, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 10.2512, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.26151222285389425, | |
| "grad_norm": 0.05340224876999855, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 10.2513, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2637862421830586, | |
| "grad_norm": 0.050805795937776566, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 10.2518, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.26606026151222284, | |
| "grad_norm": 0.052861958742141724, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 10.2513, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.26833428084138716, | |
| "grad_norm": 0.051216233521699905, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 10.2499, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.27060830017055143, | |
| "grad_norm": 0.04978267103433609, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 10.2516, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.27288231949971575, | |
| "grad_norm": 0.045428961515426636, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 10.2504, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27515633882888, | |
| "grad_norm": 0.043029140681028366, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 10.2521, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.27743035815804434, | |
| "grad_norm": 0.050259657204151154, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 10.2512, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.27970437748720867, | |
| "grad_norm": 0.05161508545279503, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 10.2507, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.28197839681637293, | |
| "grad_norm": 0.05742611736059189, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 10.2495, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.28425241614553726, | |
| "grad_norm": 0.05515799671411514, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 10.2499, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2865264354747015, | |
| "grad_norm": 0.045543234795331955, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 10.2516, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.28880045480386585, | |
| "grad_norm": 0.049780476838350296, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 10.2558, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.2910744741330301, | |
| "grad_norm": 0.054533567279577255, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 10.2503, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.29334849346219444, | |
| "grad_norm": 0.059958189725875854, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 10.251, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.2956225127913587, | |
| "grad_norm": 0.05659060552716255, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 10.252, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.297896532120523, | |
| "grad_norm": 0.04968244209885597, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 10.2512, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3001705514496873, | |
| "grad_norm": 0.05463722348213196, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 10.2527, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3024445707788516, | |
| "grad_norm": 0.055696289986371994, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 10.2511, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.30471859010801594, | |
| "grad_norm": 0.05913258343935013, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 10.2519, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3069926094371802, | |
| "grad_norm": 0.059174925088882446, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 10.2557, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30926662876634453, | |
| "grad_norm": 0.0705358237028122, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 10.254, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.3115406480955088, | |
| "grad_norm": 0.06222542002797127, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 10.2551, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3138146674246731, | |
| "grad_norm": 0.06666737794876099, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 10.2538, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3160886867538374, | |
| "grad_norm": 0.06568934768438339, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 10.2526, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3183627060830017, | |
| "grad_norm": 0.07190611958503723, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 10.254, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.320636725412166, | |
| "grad_norm": 0.08045150339603424, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 10.2538, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.3229107447413303, | |
| "grad_norm": 0.07119625061750412, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 10.2551, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.3251847640704946, | |
| "grad_norm": 0.0785309299826622, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 10.2621, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.3274587833996589, | |
| "grad_norm": 0.08730926364660263, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 10.2574, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3297328027288232, | |
| "grad_norm": 0.11653955280780792, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 10.2599, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.3320068220579875, | |
| "grad_norm": 0.1224011480808258, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 10.2629, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.3342808413871518, | |
| "grad_norm": 0.09675507247447968, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 10.2679, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.3365548607163161, | |
| "grad_norm": 0.15256762504577637, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 10.2585, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.3388288800454804, | |
| "grad_norm": 0.13102629780769348, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 10.2567, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.34110289937464466, | |
| "grad_norm": 0.19902172684669495, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 10.2625, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34110289937464466, | |
| "eval_loss": 10.252975463867188, | |
| "eval_runtime": 2.1631, | |
| "eval_samples_per_second": 342.56, | |
| "eval_steps_per_second": 85.987, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.343376918703809, | |
| "grad_norm": 0.052708208560943604, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 10.2501, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.34565093803297325, | |
| "grad_norm": 0.057368677109479904, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 10.2502, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.3479249573621376, | |
| "grad_norm": 0.0579616017639637, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 10.2508, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.3501989766913019, | |
| "grad_norm": 0.05553217604756355, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 10.2497, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.35247299602046617, | |
| "grad_norm": 0.06146073713898659, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 10.2516, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3547470153496305, | |
| "grad_norm": 0.04849397763609886, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 10.2505, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.35702103467879476, | |
| "grad_norm": 0.06262645870447159, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 10.2494, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3592950540079591, | |
| "grad_norm": 0.05687860772013664, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 10.2502, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.36156907333712335, | |
| "grad_norm": 0.06951886415481567, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 10.248, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.36384309266628767, | |
| "grad_norm": 0.05946781858801842, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 10.2498, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36611711199545194, | |
| "grad_norm": 0.057177819311618805, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 10.2488, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.36839113132461626, | |
| "grad_norm": 0.07202157378196716, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 10.2495, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3706651506537806, | |
| "grad_norm": 0.06073342636227608, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 10.2483, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.37293916998294485, | |
| "grad_norm": 0.0646699070930481, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 10.2485, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.3752131893121092, | |
| "grad_norm": 0.06538694351911545, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 10.2482, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37748720864127344, | |
| "grad_norm": 0.05704918131232262, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 10.2491, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.37976122797043776, | |
| "grad_norm": 0.06896015256643295, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 10.2474, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.38203524729960203, | |
| "grad_norm": 0.06041896715760231, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 10.2488, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.38430926662876636, | |
| "grad_norm": 0.0520286001265049, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 10.2479, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3865832859579306, | |
| "grad_norm": 0.07203182578086853, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 10.2488, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38885730528709495, | |
| "grad_norm": 0.05736120045185089, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 10.2492, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3911313246162592, | |
| "grad_norm": 0.05721374601125717, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 10.2485, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.39340534394542354, | |
| "grad_norm": 0.050962936133146286, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 10.2492, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.39567936327458786, | |
| "grad_norm": 0.058926161378622055, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 10.2488, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3979533826037521, | |
| "grad_norm": 0.05292151868343353, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 10.2493, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.40022740193291645, | |
| "grad_norm": 0.05098384991288185, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 10.2502, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.4025014212620807, | |
| "grad_norm": 0.05969351902604103, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 10.2525, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.40477544059124504, | |
| "grad_norm": 0.04939919710159302, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 10.2493, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.4070494599204093, | |
| "grad_norm": 0.0583348385989666, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 10.2516, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.40932347924957363, | |
| "grad_norm": 0.06072559207677841, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 10.2488, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4115974985787379, | |
| "grad_norm": 0.057481151074171066, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 10.2494, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.4138715179079022, | |
| "grad_norm": 0.054160043597221375, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 10.2507, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.4161455372370665, | |
| "grad_norm": 0.05530329793691635, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 10.2524, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.4184195565662308, | |
| "grad_norm": 0.050927530974149704, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 10.2509, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.42069357589539513, | |
| "grad_norm": 0.05146843567490578, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 10.2522, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4229675952245594, | |
| "grad_norm": 0.04710632190108299, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 10.2521, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.4252416145537237, | |
| "grad_norm": 0.06215859204530716, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 10.2529, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.427515633882888, | |
| "grad_norm": 0.049723975360393524, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 10.2533, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.4297896532120523, | |
| "grad_norm": 0.050091955810785294, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 10.254, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.4320636725412166, | |
| "grad_norm": 0.06934883445501328, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 10.2515, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4343376918703809, | |
| "grad_norm": 0.07753632962703705, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 10.2538, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.43661171119954517, | |
| "grad_norm": 0.0843120664358139, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 10.2557, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.4388857305287095, | |
| "grad_norm": 0.07512934505939484, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 10.2539, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.4411597498578738, | |
| "grad_norm": 0.10326453298330307, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 10.2561, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.4434337691870381, | |
| "grad_norm": 0.07381079345941544, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 10.2549, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4457077885162024, | |
| "grad_norm": 0.11267664283514023, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 10.2599, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.4479818078453667, | |
| "grad_norm": 0.09136492758989334, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 10.262, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.450255827174531, | |
| "grad_norm": 0.11963188648223877, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 10.2693, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.45252984650369527, | |
| "grad_norm": 0.1522008627653122, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 10.262, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.4548038658328596, | |
| "grad_norm": 0.14870856702327728, | |
| "learning_rate": 0.0, | |
| "loss": 10.2658, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4548038658328596, | |
| "eval_loss": 10.252269744873047, | |
| "eval_runtime": 2.1662, | |
| "eval_samples_per_second": 342.072, | |
| "eval_steps_per_second": 85.864, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 31262072635392.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |